Photodetecors fabricated with mosaic silicon wafers can be approximately categorized into two kinds: pn-junction photodetectors and charge-coupled device (CCD) photodetectors. There are two structures of pn-junction photodetectors: photodiode and phototransistor shown respectively in FIG. 1A and FIG. 2A. The phototransistor structure is comprised of a photodiode connected to the base terminal of a bipolar transistor for the capability of charge amplification. By incorporating a metal oxide semiconductor (MOS) transistor switch in any of the pn-junction photodetectors shown in FIG. 3A and FIG. 4A, a charge-integration mode photodetector element can be implemented on the integrated circuits (ICs) with improved photoresponse sensitivity. This also facilitates the integration for a large image sensing array.
The use of a phototransistor as a photodetector that utilizes photocurrent amplification to increase photosensitivity was first mentioned by William Schockley et. al., in a paper entitled "p-n Junction Transistor," Phys. Rev., 83,151 (1951). A charge integration mode phototransistor photodetector that was configured as an image capture device was described in a the paper by Rudy Dyck and Gene Weckler entitled "Integrated Array of Silicon Photodetectors for Image Sensing," IEEE Transactions On Electron Devices, Vol ED-15, No.4, April 1968.
A contact image sensor (CIS) module, that uses a charge integration mode phototransistor as the photodetector element, was presented in a paper by E. E. Anderson and Weng-Lyang Wang entitled "A Novel Contact Image Sensor (CIS) Module for Compact and Lightweight Full Page Scanner Applications," SPIE Vol. 1901 Cameras, Scanners, and Image Acquisition Systems (1993), pages 173-181. A contact-type, color image sensor using color phototransistors was presented in a paper by Tadahiko Hamaguchi, et.al., entitled "Contact-type Color Image Sensor Using Color Phototransistors," SPIE Vol. 2172, pages 167-174.
Contact image sensor modules currently used in black-and-white fax machines are built with charge-integration phototransistor sensing elements of FIG. 4A, because of the simplicity that is a result of the self-resetting capability of the phototransistor during the readout period. However, these CIS scanners have limited dynamic range for image reproduction, and are unsuitable for gray level image scanning applications. The limited dynamic range is the result of being operated in the non-linear photoresponse transfer function region and the after-image effect (or image lag) caused by the incomplete and non-linear self-resetting process of the phototransistor.
A substantial improvement in contact image scanners over those using charge-integration phototransistor sensing elements is described in co-pending patent application Ser. No. 08/654,394 filed May 28,1996 by Pao-Jung Chen, titled "Charge Integration Photodetector Having A Prechargeable Charge Storage Node".
By inserting certain DC-biasing charge through a precharge switch to the base terminal of the npn transistor prior to the charge integration process, an improved charge-integration phototransistor (as shown in FIG. 5A) eliminates the non-linear and after-image problems of the conventional charge-integration phototransistor shown in FIG. 4A, and reaches the performance of comparable CCD image sensors. However, the process technologies required in the base-biased phototransistors are not fully compatible with the existing foundries' standard CMOS process. Also, the subordinate circuits for operating this phototransistor image sensors are cumbersome and power consuming.
The other image sensor built with voltage-pickoff charge-integration photodiode sensing elements shown in FIG. 6A, has been used in image sensing applications since the early 1970's. This sensing element consists of a resettable reverse-biased photodiode connected to the high impedance gate terminal of a MOS transistor operating as a source-follower amplifier. The source follower configuration provides the low output impedance needed to drive the external readout circuits.
The concept of operation, circuit configurations, and especially the types of buffered transistors used for charge-integration photodiodes and for charge-integration phototransistors are quite different. In the operation of the voltage-pickoff photodetector, the photodiode and the gate terminal of the source-follower transistor are initially reset to a DC voltage, usually to one of the power buses. This causes the source-follower transistor to operate in the top end of the active region under the dark level condition. When the light illumination increases, the source-follower transistor operates toward the cut-off region. However, the operation of the charge-integration phototransistor is just the opposite. The photodiode and the base terminal of the npn transistor are reset to a voltage near the cut-off region under the dark level condition. When the light illumination increases, the emitter-follower transistor operates toward the top end of the active region. Their differences are illustrated in the circuit diagrams in FIGS. 5A and 6A, and in the photoresponse transfer characteristics in FIGS. 5C and 6C respectively. Both photodetctors are fabricated with CMOS technology on the same type substrate. However, the transistor body effect of the reset switch and the readout switch of the voltage-pickoff photodetector complicate the circuit design and process technology, and reduce the dynamic range and readout rate of the image sensing devices that use voltage-pickoff charge-integration photodiodes as sensing elements.
Integrating photodiode area-array image sensors with signal processing circuits on one chip using CMOS technology is currently under heavy development for emerging multimedia applications. Presentations on this topic were included in the IEEE ISSCC conferences of '96 and '97.